Unsupervised Multi-Sensor Condition Monitoring of AODD Pump Systems Using Physics-Informed Health Indices and Gaussian Mixture Models

Cold-Start Motor Condition Monitoring Using Multi-Sensor Signals and Pseudo-Supervised Machine Learning.

As predictive maintenance transitions from the data-centric paradigm of Industry 4.0 to the sustainable, human-centric framework of Industry 5.0, diagnosing servo motor conditions faces the dual challenges of data scarcity and a profound lack of labeled fault samples. To address this cold-start problem, we present a pseudo-supervised machine learning framework evaluated on a custom five-channel dataset comprising 199 servo motor telemetry samples (current, voltage, temperature, humidity, and vibration). The methodology integrates hard structural partitioning (k-means) and soft posterior confidence estimation (Gaussian Mixture Models) to characterize operating modes without prior annotation. Concurrently, an Isolation Forest model quantifies anomaly intensity and establishes a dynamic quantile-based threshold. A critical innovation of this research is the deterministic risk mapping derived from engineering priors; it defines the "high-risk" (abnormal) state by inversely weighting the physical safety margins of the sensors. This mechanism strictly maps unsupervised clusters to binary pseudo-labels. These labels are subsequently used to supervise downstream discriminators (Random Forest and Support Vector Machine). The final online diagnostic outputs a score-level fusion of the classifier probability and the GMM posterior, gated by the anomaly threshold. Quantitative evaluation demonstrates that the Random Forest model achieved a perfect F1 score of 1.000, while the comparative SVM yielded an F1 score of 0.997, proving the framework to be a robust, interpretable, and highly accurate solution for cold-start industrial health monitoring.

Unsupervised Predictive Maintenance for Multivariate Pump Sensor Streams Using Isolation Forest: Health Score Calibration and Persistence Filtering

Physics-Informed Machine Learning and Multi-Sensor Fusion for Structural Health Monitoring: A Bridge Case Study

Ensuring the integrity of critical infrastructure, such as bridges, dams, and large-scale structures, is essential to safety, reliability, and operational continuity. These assets are exposed to mechanical, thermal, environmental, and operational loads that can accelerate fatigue, corrosion, wear, and other deterioration mechanisms. Traditional inspections are periodic and costly, while sensor-based monitoring systems generate large volumes of imperfect field data that are difficult to interpret without engineering context. This paper presents a physics-informed machine learning and multi-sensor fusion framework for fatigue-oriented Structural Health Monitoring (SHM) and predictive maintenance. The method is demonstrated on an instrumented steel-concrete composite bridge using resistive strain, fiber Bragg grating (FBG) strain and temperature, and displacement measurements. In decision-critical SHM applications, uncertainty arises from noisy sensor signals, thermal effects, drift, missing data, and ambiguous operating events. The proposed framework reduces this uncertainty by applying physics and engineering principles to structure the correction, validation, and interpretation of sensor data. Machine learning supports thermal compensation, anomaly screening, cross-sensor substitution, and confidence tagging, while deterministic engineering methods remain responsible for strain interpretation, event extraction, rainflow-style cycle counting, and fatigue damage indicators. Results show that multi-year monitoring data can be reduced into compact fatigue-relevant features while preserving traceability to raw measurements. A supervisory agentic layer coordinates data-quality checks, multi-sensor consistency review, and confidence-tagged substitution, creating an auditable workflow for engineering decision support. SHM value does not come only from better sensors or better ML models. It comes from traceable workflows that turn imperfect field data into reliable, reviewable, and actionable engineering evidence.

Multi-Sensor Fusion and Frequency-Domain Analysis for Predictive Maintenance of Industrial Induction Motors

Unplanned failures of induction motors impose serious operational and financial penalties on industrial facilities, yet the fault signatures that precede such failures are detectable well in advance through careful sensor instrumentation and data-driven analysis. This paper presents an end-to-end Internet-of-Things (IoT) predictive maintenance scheme based on two off-the-shelf sensors: a DS18B20 one-wire digital thermometer and 2 piezo vibration sensor modules, with an ESP32 edge device for running the full machine learning pipeline offline, independent from any cloud services. Four operating scenarios are considered: healthy condition, BPFO (bearing outer race fault), misaligned shaft, and rotor imbalance. Based on 1-second sampling intervals, 17 descriptors are derived, including statistics in the time domain, Fourier harmonic peaks, energy ratios between different frequency bands, and temperature gradients measured across all sensors. A dual-stage feature selection method using mutual information (MI) score and Random Forest mean decrease impurity (MDI) ranking reduces the number of features to the 10 most relevant descriptors, reducing the computational complexity by 41% at the expense of 5.6% F1-macro. On a balanced 600-sample synthetic dataset, the resulting Random Forest classifier attains 87.3% hold-out accuracy, 91.0±1.9% five-fold cross-validation accuracy, and a macro area-under-the-ROC-curve of 0.980. End-to-end inference takes just 39 ms on the ESP32, easily meeting the 200 ms requirement for real-time alerting.

Intelligent structural health monitoring using convolutional neural networks and IoT sensor fusion

Structural health monitoring (SHM) is essential for ensuring the safety and longevity of critical civil infrastructure such as bridges, buildings, and dams. Traditional SHM approaches rely heavily on manual inspection and threshold-based alarm systems, which are prone to high false alarm rates and limited sensitivity to incipient damage. This paper proposes an intelligent SHM framework that integrates one-dimensional Convolutional Neural Networks (1D-CNN) with Long Short-Term Memory (LSTM) networks for automated damage detection from vibration sensor data acquired through Internet of Things (IoT) sensor networks. The proposed architecture employs a dual-branch feature extraction strategy: the 1D-CNN branch captures local spatial patterns from raw acceleration signals, while the LSTM branch models temporal dependencies across sequential sensor readings. An attention-based sensor fusion module aggregates information from heterogeneous sensor types including accelerometers, strain gauges, and temperature sensors, enabling comprehensive structural state assessment. Transfer learning is applied to adapt models pre-trained on large-scale simulated datasets to real-world bridge monitoring scenarios. Extensive experiments on the LANL structural damage detection dataset, the Z24 bridge benchmark, and a custom simulated structural dataset demonstrate that the proposed method achieves damage detection accuracy of 96.5%, significantly outperforming conventional machine learning baselines including SVM (84.1%), Random Forest (85.5%), and standard MLP (87.2%), while maintaining a false alarm rate below 3.2%. Ablation studies confirm the contribution of each architectural component to the overall performance.

AutoML-Optimized Vision Transformers for Multimodal Structural Health Monitoring from Multi-Channel Feature Tensors

Reliable structural health monitoring (SHM) of wind turbines requires methods capable of handling high-dimensional, heterogeneous multi-sensor data under both labeled and label-scarce conditions, while remaining computationally efficient for real-world deployment. This paper presents a unified SHM framework that integrates feature-level data fusion, Vision Transformer (ViT) classification, multi-objective AutoML optimization, and unsupervised anomaly detection within a single pipeline. Multi-channel signals from 26 sensors are preprocessed and encoded into a compact three-channel feature tensor combining statistical descriptors, spectral and wavelet features, and a PCA-denoised inter-sensor correlation matrix. This representation enables efficient storage of time-window information and captures both local signal characteristics and global cross-sensor dependencies. A lightweight ViT is trained for supervised fault classification, while its architecture is optimized using NSGA-II to jointly maximize predictive performance and minimize computational cost, enabling deployment on resource-constrained edge devices. To address label scarcity, an autoencoder trained solely on normal-condition data is used for anomaly detection. Comparative evaluation of reconstruction- and latent-space-based metrics shows that Mahalanobis distance in the latent space provides superior sensitivity to subtle faults. Validation on the ETH Aventa AV-7 dataset demonstrates up to 98.4\% macro-F1 in classification and robust anomaly detection performance. The results confirm that the proposed multisensor fusion strategy provides a reliable and scalable pipeline for SHM of real structures, applicable to both supervised and unsupervised scenarios under practical computational constraints.